Most cases of PA are metastatic at presentation. Metastatic PA is a rapidly progressive and inevitably fatal disease with an average survival of less than 6 months.⁴ Improvements in the technical aspects of surgical management, advances in adjuvant treatment and regionalization of treatment have all played roles in improving long-term survival of patients treated with curative intent.^5,6 Owing to a high locoregional recurrence rate, adjuvant chemotherapy should be considered in all patients after curative resection, regardless of margin and nodal status,^7,8 and patients with incurable disease may be offered palliative chemotherapy.⁸

Owing to the high fatality rate, epidemiologic studies of PA are challenging and limited by difficulties contacting living patients.⁹ Since PA is rapidly fatal (fatality rate = 95% at 5 yr), it is of great importance to identify and contact all patients shortly after diagnosis. An innovative approach to rapidly identify patients with PA across the province of Ontario, the electronic pathology reporting system (E-path), was implemented to provide the fastest source of cancer information. The Ontario Cancer Registry (OCR) uses the Pathology Information Management System (PIMS), which relies on E-path and supports rapid computerized case ascertainment.¹⁰ The E-path system is a centralized database used to collect electronic pathology information from Ontario laboratories processing tumour specimens. E-path not only provides more timely pathology reporting compared with paper pathology reports, but also has been shown to increase the completeness of reporting. This is especially important in studies of patients with a rapidly progressive disease such as PA. In the E-path system, pathology reports arrive from each laboratory electronically and are queued in a database for on-screen review by health record technicians. This electronic transfer of reports occurs daily for most of the laboratories, or weekly for some of the smaller-volume laboratories. If the health record technicians determine that a pathology report contains reportable findings, the report is coded and consolidated with the OCR database. Reports of specific cancers, such as PA, are filtered and printed for review by study personnel.¹⁰

Little is known about the system-wide management of highly fatal cancers such as PA. Although administrative databases, such as the OCR, Canadian Institute for Health Information (CIHI) Discharge Abstract Database and others, are useful in providing information on patients with cancer in Ontario, more sophisticated, population-based efforts are necessary to study the epidemiology, management and outcome of PA in greater detail. The Ontario Pancreas Cancer Study (OPCS) began on Apr. 1, 2003, and is an ongoing prospective case–control study. The primary objective of the OPCS is to collect data about genetic and epidemiologic factors involved in the etiology of PA. The OPCS is also 1 of 7 data collection sites that make up a multicentre, multidisciplinary Pancreatic Cancer Genetic Epidemiology (PACGENE) Consortium based out of the Mayo Clinic in Rochester, Minnesota, that aims to identify susceptibility genes in high-risk familial pancreatic cancer (FPC) kindreds.¹¹

The goals of the present study were to use data from the OPCS to characterize the management of PA in Ontario and to describe the use of a survey to supplement cancer registry data.

We asked the physician identified on the pathology report of each patient for confirmation of the PA diagnosis and for permission to contact the patient. Consent was obtained for each patient. Passive consent was assumed for patients with known PA if the physician had not replied after 2 weeks. In cases of ambiguous pathology reports, we did not contact the patients without first obtaining clarification of the diagnosis from the physician.

Patients were mailed a package including an introductory letter; consent forms; a family history questionnaire (FHQ) designed to obtain information about the family history of cancer, diabetes and pancreatitis; a personal history questionnaire (PHQ), including questions on environmental and epidemiologic exposures, such as medical history, medication use, dietary patterns, physical activity, reproductive history, chemical exposures and alcohol and tobacco consumption; and a clinical patient questionnaire (CPQ) that addressed treatment issues and knowledge of clinical trials in PA. A release form was included for patients to complete, allowing us to obtain medical records describing their surgical, adjuvant and palliative treatments. Within 2 weeks of the questionnaire mailing, we sent a follow-up postcard to patients who had not returned the package. A telephone call was made 2 weeks later to individuals who had not responded, and a second questionnaire package was mailed on request.

On receipt of the medical records, we reviewed the notes to extract data on specimen pathology (e.g., specimen type, histology, differentiation, regional lymph nodes, number of nodes removed, number of positive nodes, tumour–node–metastasis [TNM] staging, size of the tumour, margins and angiolymphatic and perineural invasion); curative procedures (e.g., distal or total resections, Whipple, pancreatectomy, vascular resection); palliative procedures (e.g., gastrojejunostomy, choledocho/hepatico-jejunostomy); chemotherapy and radiation treatments; and recurrence of PA (e.g., site, date of recurrence, pathology obtained).

We defined surgery with curative intent as any resection intended to remove the entire tumour with pathologically negative margins; this included both completed and aborted resections. We defined aborted surgery with curative intent as an intraoperative decision that the tumour was unresectable with no attempt to remove the tumour in its entirety. Completed surgery with curative intent included all operations in which a removal of the tumour was successful; pathologically, these resections may have had either positive or negative margins.

Academic institutions included all major hospitals affiliated with the 5 general surgery residency programs in the province (e.g., Hamilton Health Sciences Centre, Sunny-brook Health Sciences Centre, University Health Network, Mount Sinai Hospital, The Ottawa Hospital, London Health Sciences Centre and Kingston General Hospital), whereas nonacademic centres were not formally affiliated with any university or residency program.

Staging of PA was conducted using the TNM classification of the Union for International Cancer Control (UICC; 6th edition, 2002).¹² The staging classification incorporated all medical reports generated within 3 months before and after the earliest pathologic diagnosis. Specific rules for consistency of staging classification were followed and are outlined in Appendix 1, available at www.cma.ca/cjs.

Table 2 shows differences in treatment among the responders (n = 351). Of the 189 (54%) responders who underwent surgery with curative intent, 157 (83%) surgeries were completed, whereas 32 (17%) were aborted and a palliative operation performed. Adjuvant or palliative chemotherapy was administered to less than half of the responding patients. Most respondents were not enrolled in any clinical trials.

A significantly greater proportion of patients aged 70 years or younger underwent surgery with curative intent than patients older than 70 years (58% v. 45%, p = 0.03; Table 3). There was no correlation between age and completion of surgery with curative intent, margin status, administration of chemotherapy or enrolment in clinical trials.

Table 4 shows the correlation of treatment site with management of PA responders. Surgeries with curative intent performed in academic centres were statistically more likely to be completed than those performed in nonacademic centres (90% v. 67%, p < 0.001). There was no statistical difference between treatment in academic or nonacademic centres with respect to margin status, enrolment in clinical trials or use of any form of chemotherapy.

At the time of diagnosis, TNM classification was used to stage all patients (Appendix 1). Treatment plans were shown to be different based on stage of disease (Table 5). All of the patients with stage-I and -II tumours underwent surgery with curative intent; these surgeries were completed in all but 1 patient with stage-II tumours. Stage-III and -IV cancers were identified based on postoperative investigations or intraoperatively; thus, 17 of 60 (28%) patients with stage-III tumours and 19 of 78 (24%) patients with stage-IV tumours underwent an attempted surgery with curative intent because metastatic disease was not evident preoperatively. None of the patients with stage-III tumours had completed curative resections; whereas 9 of 19 (47%) patients with stage-IV tumours had completed curative resections as metastases became evident within 3 months postoperatively. Of note, completion of surgery with curative intent only applies to stage-IV tumours that were diagnosed postoperatively as, by definition, stage-III tumours were unresectable owing to invasion into adjacent structures. Patients with stage-I tumours received chemotherapy less often than patients with stage-II, -III or -IV tumours (p < 0.001). When stratified by both treatment site and stage, chemotherapy use was significantly different between academic (45%) and nonacademic centres (22%) for patients with stage-II PA (p = 0.05).

The OPCS was designed as a population-based registry of incident cases of PA in Ontario for epidemiologic, genetic and health services and outcomes research. Even with the use of E-path reporting and rapid case ascertainment, the overall response rate in the OPCS has never exceeded 35%, which is consistent with other attempts to obtain data through population-based registries of PA with the use of questionnaires.¹⁴ Despite the low response rate, we sought to characterize responders and nonresponders and use the OPCS to describe the management of PA in our registry of patients in Ontario. Our experience and observations underscore the need for further prospective, population-based studies of PA with emphasis on data collection from patients, physicians and existing administrative databases in the province. Our study simultaneously highlights the practical obstacles to progress, such as the rapid identification of patients with newly diagnosed PA.

We found that nonresponders were significantly older and more likely to have received treatment in a nonacademic centre than responders. This inverse relation between age and enrolment in prospective studies has been previously shown in cancer trials.¹⁵ Increasing age has been correlated with less aggressive management, which has been shown to be a barrier to participation in these studies owing to patient apathy and a sense of helplessness.¹⁶ Patients treated in nonacademic centres were less likely to respond to our recruitment strategy, possibly demonstrating the potential influence of an academic environment on the attitudes of these patients. In the province of Ontario, academic centres are geographically situated in the larger urban centres, and it has been shown that rural patients are less likely to participate in cancer research.¹⁶ However, it is also plausible that patients who are more likely to participate in research studies may be more likely to initially seek medical attention at an academic centre.

In our registry, patients aged 70 years and younger were more likely to undergo surgery with curative intent than those older than 70 years. Older patients are more likely to have medical comorbidities that would preclude surgery, and this was not accounted for in this study. However, the small number of patients unfit for the anesthetic or operative risk would likely not account for the statistical significance between these groups. Such a difference may be owing to physician education and attitudes toward aggressive surgical treatment in elderly patients and a tendency for physicians to be more aggressive in the treatment of younger patients.¹⁷ Other studies have also reported that older patients refuse pancreatic surgery more often than younger patients;¹⁸ however, this may be owing in part to the reluctance of surgeons to operate on older patients.

There is an ongoing trend toward centralization of management of PA in high-volume institutions owing to lower perioperative surgical morbidity and mortality.^6,19,20 Our study examined the differences in the management of PA between academic and nonacademic centres and revealed that, within our registry, the number of aborted surgeries with curative intent was significantly higher in nonacademic centres than academic centres. However, when surgery with curative intent was completed, there was no significant difference in the percentage of positive margins between these 2 groups. This suggests that, although the decision to abort versus complete curative resection may have varied between academic and nonacademic centres, once the decision was made to complete a curative resection, surgical technique to acquire negative margins did not vary. This may indicate that a more aggressive approach to completion of surgery with curative intent is taken in academic centres with a higher threshold for aborting an operation with curative intent. However, it is also possible that the quality of preoperative staging evaluation and radiographic studies (e.g., pancreas protocol computed tomography, endoscopic ultrasonography) may be higher at academic centres; as a result, patients with technically unresectable disease may be identified preoperatively more often at academic centres, lowering the likelihood that an operative procedure would need to be aborted based on unexpected intraoperative findings.

All patients with stage-I or -II tumours for whom data was available underwent operations with curative intent, and all but 1 operation was completed. Bilimoria and colleagues¹⁸ found that 38.2% of their patients with stage-I or -II PA with no identifiable contraindication to surgery did not undergo surgical management. The likely difference in our database, in which all responders with stage-I or -II disease underwent surgical resection, is that our case ascertainment was limited to pathologic diagnosis and to questionnaire responders. As the usual management of clinically diagnosed early-stage PA is resection with no preoperative biopsy, the only pathology for these patients, which was our means of case ascertainment, was their surgical resection specimen. Owing to incomplete diagnosis of stage preoperatively, 36 (26%) patients with stage-III or -IV tumours had surgery with curative intent. There was no significant difference between the ability to diagnose unresectable tumours in academic compared with nonacademic centres. All surgeries with curative intent in patients with stage-III tumours were aborted in both academic and nonacademic centres, whereas the intraoperative decision to abort these surgeries in patients with stage-IV tumours was made more frequently in nonacademic centres than academic centres.

It has recently been shown that adjuvant chemotherapy should be offered to all patients with a diagnosis of PA.⁸ Although it is difficult to study the proportion of patients being offered chemotherapy, our analysis reveals that less than half of our patients received any adjuvant or palliative treatment, with no statistical differences demonstrated by either age or treatment site. This may be owing to our data collection having begun in 2003 before these guidelines were established. As stated, patient choice may also play a role; however, it is difficult to assess the proportion of patients being offered treatment and declining it. There was a significant difference in the use of chemotherapy based on stage of disease; patients undergoing curative surgical treatment for stage-I or -II tumours were less likely to receive chemotherapy as an adjunct to their treatment (13% and 40%, respectively) than those with stage-III or -IV tumours. Once again, this may be owing to both physician and patient knowledge and attitudes toward the role of chemotherapy.

Another finding in our study was the low enrolment in clinical trials across all patients, with no correlation between age or stage of disease and likelihood of enrolment. This may not be surprising, considering that most patients with cancer do not participate in clinical trials, and many patients with PA have poor performance status, limiting their eligibility for trials.

There are several limitations to our study. As PA is rapidly fatal, survival bias almost certainly affected our results. Although we used rapid case ascertainment to recruit patients, 44% of the patients died before enrolment. In addition, our response rate never exceeded 35%. Nonresponders may have had more aggressive disease and may have been less responsive to treatment. As a result, patients who underwent surgery may have been more eager to participate than patients with inoperable tumours. In fact, over 40% of our patients underwent surgery, which is double the number seen in the general PA population.^2,3,11 Selection bias is also possible because we only recruited individuals with a pathologically confirmed diagnosis of PA. Over 50% (59% in 2002, Cancer Care Ontario: personal communication, 2006) of patients with PA in Ontario never have tissue confirmation while they are alive, and are only reported to the OCR from hospital records or death certificates.

Some patients may have been treated at more than 1 hospital and, although we used pathology reports and operative notes to classify the treatment centres, other decisions and treatment plans may have occurred elsewhere and were not accounted for in the analyses. Hospital records on chemotherapy were not always complete, although patient-acquired history was also used to determine use of chemotherapy. Of the 351 responders, only 269 had complete data from which to determine the stage of disease (i.e., in some patients a full work-up was not complete after diagnosis). Another limitation was the variation of synoptic pathology reporting, with different laboratories having varying thresholds for describing positive and negative margins.